Control of a nonlinear CSTR model by gain scheduling of controller tuning

Nonlinear processes can cause a controller to become oscillatory in some situations, sluggish under other conditions. Gain scheduling is a method to improve control of a nonlinear process. In some cases, large changes in disturbances can cause the process to become unstable, while equally large changes, but in opposite sign, can result in sluggish behavior.; In this thesis, conventional PI control with a constant gain will be compared to gain scheduling of a PI controller for a nonlinear process. The model used will be a nonisothermal CSTR governed by the Arrenhius equation. A Maple program, which produces responses by numerical methods and incorporates PI control, will be utilized for the comparison.; Gain scheduling of the PI controller produced better responses than those by traditional PI control with a constant gain. Integral Absolute Error (IAE) and the Integral Square Error (ISE) calculations were employed to compare the two results. ISE and IAE values were found for eight simulations, resulting in only one ISE value being more favorable for traditional PI control with a constant gain.; The next logical step for this study is to compare gain scheduling control to Nonlinear Model Predictive Control (NMPC). A neural network could be used as the process model in the NMPC algorithm and can be an excellent representative of the input-output relationship for a system. A section on NMPC and neural networks is included for information only in the theory section.